# Amendment 2026-09-22 — a locked forecast, and a slate candidate outside the denominator

Per `PRE-REGISTRATION.md` §8: a dated amendment with a stated reason and a flag for whether it affects a locked prediction.

**Affects a locked prediction:** none existing. This amendment *creates* one — `PRED-2026-09-22-PRESS-ACCESS` in `slate.json` — and adds `PRESS-ACCESS` to the slate as a candidate with no band prediction, which keeps it out of the breadth denominator by the guard in `pipeline/breadth.py`.

**Locked by:** Briar Greenway, 2026-09-21 21:54 PDT (2026-09-22T04:54:53Z), by replying "lock" to the drafted operative form. **Drafted by:** Claude. The analyst's statement is recorded verbatim in the slate; the three operative predictions P1–P3 and their triggers are Claude's rendering of it, which he locked as his.

---

## §1 — What was locked

The analyst's statement, verbatim: *"my prediction for this is that it will also not be fixed cheaply. and it most likely won't be fixed at all. leaving the space required for a purer propagandist state media to replace it. that combined with the media take over and mergers by the tech oligarchs could really distort any truth by never letting anything but lies be broadcast."*

The operative form, with the trigger that falsifies each:

| | Claim | Falsified if | Check on |
|---|---|---|---|
| **P1** | Not fixed cheaply: no credential restored by voluntary act before a court order. | Any of the three outlets' credentials restored without a court compelling it. | 2026-12-21 (T+90, past the midterms) |
| **P2** | Not fixed in fact: by T+365 either no final appellate ruling restores access, or a ruling exists and independent TV-pool coverage of presidential events is still below its 2026-09-18 baseline. | The five-network pool is back to baseline under a complied-with order. | 2027-09-22 |
| **P3** | Substitution, not vacuum: by T+365 the share of presidential-event footage from administration feeds or single-outlet friendly pools exceeds the five-network pool's share, **and** at least one state-imposed editorial board at CNN or CBS has been altered, overridden or allowed to lapse. | Pool share recovers to majority, **or** both boards are intact and demonstrably exercised at T+365. | 2027-09-22 |

**Not locked, on purpose:** "never letting anything but lies be broadcast." An absolute has no falsification trigger and would void the row under a hostile reader. Its measurable residue is P3.

## §2 — Why this is a forecast and not a score

The register locks predictions *before* scoring so that a result can embarrass the analyst. A forecast with dated triggers is the same discipline applied to the world rather than to the instrument: the analyst says in advance what would prove him wrong, and the date on which it is checked. P1 is cheap to check and checks first. P2 and P3 need two baselines that **do not yet exist** and must be recorded now or the predictions cannot be scored later:

1. The independent TV-pool share of presidential-event footage in the fortnight before 2026-09-18. Without it, P2 and P3 have no denominator.
2. The composition and charter of the CNN and CBS editorial boards at merger close, expected by 2026-09-30. Without it, "altered, overridden or lapsed" has nothing to be measured against.

Both are listed in the slate record as `baseline_measurements_needed_now`. A forecast whose baselines are never recorded is a forecast that can never lose, which is the one property the register refuses.

## §3 — The base rate the forecast rests on, and its size

Two prior instances in the same locus, both unremedied:

- **Pentagon, 2025-10-15.** Sign-or-leave pledge; nearly every outlet left; badges confiscated; OAN signed; ~15 reporters remain. Policy unchanged at eleven months.
- **AP v. Budowich, filed 2025-02.** Injunction 2025-04-08; the White House barred AP from the Oval anyway and enforcement was denied; the D.C. Circuit on 2025-06-06 confined relief to the East Room on the reasoning that the Oval Office and Air Force One are "not public"; en banc denied 2025-07-22; argued 2025-11-24; undecided at nineteen months.

That is n=2. Stated plainly so the forecast's confidence is read at the size of its base. The analyst's "most likely won't be fixed at all" is P2, and P2 is the claim the base rate supports most directly: in both prior instances the cheap fix was available, ordered or demanded, and not delivered in fact.

## §4 — The slate entry, and what keeps it honest

`PRESS-ACCESS` enters `slate.json` with role `seed_candidate` and **no band prediction**. The analyst locked a dimension-level expectation — high S3, high S5, low S1 — and no composite band. The breadth guard refuses to count a member without a locked band, so the candidate sits outside the denominator until he supplies one. That is the correct state, not a gap: a band supplied after the annex was assembled would be a prediction shaped by its evidence.

Locus: one, `LOCUS-PRESS-ACCESS`, collapsed per the codebook's coarser-is-conservative rule. Splitting the Department of Defense action from the Executive Office actions is arguable and would add a locus; it was not taken, and the annex says so.

## §5 — Objection, recorded by the drafter

The forecast's strongest form (P2) and its weakest form (the analyst's absolute) share a mechanism but not a scale, and a reader who agrees with P2 may still find P3 too strong. P3 was kept because it is the only place the consolidation claim becomes measurable inside a year; if it fails while P1 and P2 hold, that is the informative outcome, and it should be published as such rather than folded into a general "the forecast held."